EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: noreply on July 21, 2020, 10:56:56 pm
-
Great new for me today – the Siglent SSG3021X arrived.
So now we can literally do a ‘head-to-head’ comparison with a less expensive rival – the RIgol DSG 800 series.
Initial Notable Differences
Size comparison is HUGE (the Siglent is huge) – see the attached photos.
Fan Noise – significantly higher level than the Rigol
Touch screen - +1 for Siglent – I like this feature – it’s very convenient compared to key entry alone
Control panel layout – Rigol is very compact because the Rigol is a small device so needs to be compact.
If you have ‘big’ fingers – then you must take care – the Siglent keypad has significantly more ‘real estate’ – so better suited for people with big thumbs.
Frequency Check
The Rigol – out of the box has excellent frequency ACCURACY and well within the STABILITY of 2ppm specification at 25 deg Celsius – within the range of 0 to 50 deg Celsius.
Interesting observation BOTH Siglent and Rigol do not provide a ‘specifications page’ inside their respective user manuals.
I guess they assume you are past this stage – where you keep referring to the specifications NOT!
I think they SHOULD ALWAYS have a specifications page INCLUDED in your purchase – yeah you will say its online – sure – but I want it with the product – because it saves me 1 microsecond fafing around on the internet – something I don’t need to do if included :P
So, after fafing around on the internet looking up the Specifications for the Siglent – I can confirm that it’s also within the Frequency STABILITY figure
– but WAY off in Frequency accuracy compared to the GPSDO 10MHz clock and also marginally worse than Rigol’s ACCURACY – also out of the box – no 'user' calibration.
Interesting to see that the Siglent - has menu function to 'tune' clock offset - nice to have - and possibly why they need it is because 'out of the box' Siglent has worse Frequency ACCURACY than the competitive Rigol.
BOTH instruments at room temperature (21 deg C) and ‘warmed-up for over 60 minutes.
Siglent Frequency Accuracy is: +0.95769Hz above the 10.000 000 00 MHz reference from GPSDO 10MHz source.
Rigol Frequency Accuracy was: +0.71581 Hz above the 10.000 000 00 MHz reference from GPSDO 10MHz source.
So it depends how much of a pencil neck you happen to be – like for like the RIGOL wins!
RF Signal Amplitude accuracy
At 10MHz and 0dBm output – the Siglent shows a level of -0.06dBm – hovering to 0.0 at times – this is extremely good. (see: siglent 0dbm.png)
BUT
When you think about it for a minute – you cannot trust this figure – because there is a N to N connection cable in the signal path.
I cannot believe that there is ZERO or close to zero (-0.06dBm) loss for the cable – even at 10MHZ this does not sound correct.
The Rigol shows -0.12 dBm at the same settings and conditions – I guess I am likely to believe the slightly higher figure better. (see: rigol 0dbm.png)
But BOTH devices have excellent attenuation level accuracy at 10MHz and 0dBm RF signal.
Harmonics
Firs we need to qualify some things as Rigol and Siglent have different ‘in spec’ calibration levels.
Rigol MAX output level in dBm within Calibration Specification is +13dBm
Siglent MAX output level in dBm within Calibration Specification is +13dBm - but device allows +17dBm without any UNCAL warning @ 10MHz
So for a like for like comparison meaningful values are up to +13dBm
I measured the respective harmonic levels at 13dBm for BOTH devices
At +13dBm – Rigol has no notable harmonics above -38.75dBm (see: rigol 13dbm harmonics.png)
At +13dBm – Siglent has notable Harmonics 1st harmonic is -19.57dBm – no notable 2nd / 3rd harmonics visible below 38.81dBm (see: siglent +13dBm harmonics.png)
I did not bother looking below this ‘floor level’ of the settings I used in the SVA
Rigol is a clear winner with regard to harmonics!
Please note - all of the above measurements and observations are 'quick' initial 'hands-on' testing - to give an indication where the respective instruments stand when comparing side-by-side.
It is not a precise 'test-lab' review - at least not at this time.
I can confirm that the Frequency Accuracy test has been performed with a 'test lab' quality frequency source - a GPSDO with a 10MHz reference.
More to come soon ... ;)
-
You must use same MIXER level instead of same Input connector level!
ETA: Sorry, it was my mistake when I "too quickly" look images and text about levels used. Even now I do not understand what make me think that you used different SA mixer levels (what is naturally very important when try characterize DUT's harmonics so that DUT harmonics are not messed too much with SA produced harmonics) .. Of course when I look now again I can not see any reason why I make this mistake.
-
You must use same MIXER level instead of same Input connector level! |O
Thanks for dropping by this thread rf-loop.
I'm not sure what you were referring to with your comment?
If it relates to one of the screen photos on the Siglent SSG - the +17dbm was not the level used in the harmonics testing (both the Siglent and Rigol were tested with +13dBm) - I just put-up this picture because the SSG allowed to 'set' the +17dBm - despite the specifications stating that only a maximum level +13dBm is acceptable without 'triggering' an UNCAL warning.
Interesting thing is that anything ABOVE +17dBm will trigger an UNCAL warning message.
You can physically 'set' output level to a max of +20dBm
The Rigol will allow +20dBm also - but it will issue a UNCAL warning for any levels above +13dBm - its specified value in the specifications.
I would have expected the Siglent to also issue an UNCAL warning at the +13dBm value as described in the specifications - instead of the +17dBm :-\
-
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1028104)
There's a problem with this SSG3021X measurement.
First harmonic doesn't meet spec as it should be 30dB+ down.
SSG3021X 10 MHz 10dB using Siglent 6GHz N-N cable connected to SVA1032X in FFT mode with settings to reveal the 4th harmonic.
-
LF mixer vs RF mixer so to compare apples with apples not some artificial oranges vs apples.
User error.
Anyways, why compare harmonics between 2 RF gens @ 10 MHz when this is way down in ARB frequencies. :-//
What makes you think that LF was used in EITHER of the devices during the testing?
10MHz was leftover from Frequency check - why not check the ARB levels as well as the GHz levels - this device is specified to work at 10MHz - so why cant it be tested there?
The day is long - so plenty of time to test the GHz
Nobody is selling apples here - we are simply trying to get some measurements - rather than play with some buttons :P
-
Thanks @noreply for your effort! It's good to see a direct comparison of the two current "decent" low-end SG contenders.
I may add in some data of an R&S SM300 soon since I got lucky (okay, it's up to be found out how much so... ;)) on putting an offer on a defective one (as it seems, the well-known PA fault hit again), so if the repair turns out successful, I should soon be a proud owner of a vector signal generator for round about 600EUR (of which almost one third went for shipping... :'(). In contrary to Rigol's and Siglent's designs, R&S went for a full-range mixer approach.
I'm very curious how well an almost twenty years old design (also rather low-cost at its time) compares to today's budgetary instruments.
-
Generators Wars! :box:
-
Hello tautech / Rob,
I want to deliver an important message – but at the same time do not want you to misinterpret or see this message as a negative or personal ‘dig’ at you.
THIS IS NOT ITS PURPOSE and I preface my message with this note especially to make it clear to you that my intentions are honourable and hopefully helpful for you and the community at EEVbog.
I have been an active member on this forum for a relatively short period of time.
During this time I enjoyed making almost 200 posts – it’s nice to be able to learn from others and at the same time share your knowledge and observations.
You must have been on this forum much longer – with almost 18600 posts.
During my short period here, I noticed that you tend to ‘pop-up’ in just about every thread.
Occasionally – you ‘chime-in’ to clarify an ‘open question’ – but most of the time you simply ‘chime-in’ with a comment or observation which has absolutely no value to the thread – but rather than help (which I am sure you want to do) – this type of uninvited ‘chime-in’ does more damage than good.
If you are still with me to this point and not getting a little ‘pissed-off’ – then that’s great – because I wanted you to remember my preface remark – my intentions are honourable and I truly believe that you are a ‘good bloke’ – who’s intentions are also honourable.
Again, this is not intended as a personal ‘dig’ at you – but an observation – from someone ‘new’ – so hopefully this fresh input will help you be more ‘in-tune’ because sometimes – after 18600 posts you can lose perspective and purpose which I think you have.
My self-imposed forum etiquette is simple.
If I find an interesting topic and decide to ‘read’ some posts, I don’t automatically have a desire or need or compulsion to make a comment. I will only make a comment if I feel it will ADD something POSITIVE to the subject under discussion.
Or, same logic, ask a question – only if it the answer will ADD something positive to the subject under discussion.
OK, now we have some self-imposed guidelines which I like to follow.
Now, let’s have a look at how ‘you appear’ to someone like me (and most likely many many others – who simply choose not to give you feedback) in the type of forum posts you make.
I started this new thread – as a point of focus for anyone interested in the Siglent SSG3021X RF Signal Generator.
I have been very fortunate to be given the opportunity to review this model – for myself – on a sale or return basis.
I could easily just spend a few days – doing my own non structured testing – a bit here , a bit there, check this, check that … until I have a good feel of the instruments performance.
If I like it – I will BUY it – the end!
But rather than go through this ‘privately’ I decided to do this ‘publicly’ – so maybe someone else on this forum could benefit – because they have not been fortunate enough in having a sale or return offer.
My review is not a ‘professional’ laboratory review – with high end test equipment stressing every aspect of the device under test – it’s far from it!
But despite this – I think it’s still going to be useful to many people – who simply want to ‘check it out’ – without the formal methodology of a professional testing lab.
So, I get the device unpacked and start my observations and testing.
At the same time I post my results in this thread – clearly stating in the end – more to come soon …
Shortly after, I am blessed with a well-respected senior forum member rf-loop who ‘chimed-in’ with a comment.
Upon reading the comment – I could not see or understand what rf-loop was trying to warn me about?
Perhaps I did something wrong in my test, or presented some data incorrectly?
Whatever it was – the fact that a respected forum member with far more experience than me (with a name rf-loop – you need to pay attention ;) ) chimed-in with a warning – made me think and check my post.
So I responded to rf-loop – trying to get some more insight in the meaning of his post – but using some etiquette at the same time and not being rude – by NOT responding with a comment like ‘what’s your problem mate?’ – instead I restated some observations and clarified some photos – just in case rf-loop misinterpreted something before making his comment to me.
Shortly after my response to rf-loop – you ‘chime-in’
BUT, instead of trying to be helpful – and using etiquette – you simply make a snide remark which in my opinion is clearly offensive and if that’s not enough, you then finish with a ‘slap on the face’ with another two remarks – more ‘slaps’ for each cheek :palm:
Just in case you can’t see what I am talking about …
LF mixer vs RF mixer so to compare apples with apples not some artificial oranges vs apples.
User error.
Anyways, why compare harmonics between 2 RF gens @ 10 MHz when this is way down in ARB frequencies.
The proper response would be …
I think rf-loop may be thinking that you are using LF input on the Rigol and RF input on the Siglent – hence – any harmonic tests may not be accurate as different mixes are in play.
But instead you ‘offend’ by stating …
compare apples with apples not some artificial oranges vs apples.
Then add another …
User error.
And if that’s not enough – yet another (for the kill) ..
Anyways, why compare harmonics between 2 RF gens @ 10 MHz when this is way down in ARB frequencies.
So, rather than come back at you– with my now angry self – and make a rude response, I choose to be polite and in a tong-in-cheek manner respond with a comment that I know will paint you into a corner – making it difficult for you to continue with more offensive comments.
What makes you think that LF was used in EITHER of the devices during the testing?
10MHz was leftover from Frequency check - why not check the ARB levels as well as the GHz levels - this device is specified to work at 10MHz - so why cant it be tested there?
The day is long - so plenty of time to test the GHz
Nobody is selling apples here - we are simply trying to get some measurements - rather than play with some buttons :P
Shortly after, you respond with …
Maybe my mistake on reading the DSG 800 display, sorry I'll butt out for now.
Please carry on.
Here is where I have to PAUSE – because – it is at this point that I can be of most help to you.
Let’s look closely at your response …
Maybe my mistake on reading the DSG 800 display,
You must of taken time – to ‘study’ all the information (which has not changed – been in front of you all the time) and now correctly identified what possibly rf-loop could have mistaken.
Here you could have clearly – pointed this out to me IN THE FIRST INSTANCE – alerting me to the possible mistake rf-loop made, but instead you choose to ‘snipe’ and try to score points!
Lets look at the second part of your response ...
sorry I'll butt out for now.
Now – you clearly see that you made a mistake – and PROPS TO YOU – you apologized ‘in small caps’ – a simple ‘sorry’
You then – before the ink has dried on the ‘sorry’ remark - make another comment …
I'll butt out for now.
Here you clearly acknowledge that you HAVE ‘butted-in’ – in a subject that possibly did not need your input i.e. you could have just let rf-loop reply (if he chooses) by himself.
And, wait for it, you state ..
for now.
:palm:
This once again , clearly indicated that YOU reserve the right to come back and make more uninvited remarks!
Finally, if the above is already not enough indication of your demeanor, you go on to say …
Please carry on.
As if nothing has happened – nothing – and most importantly it’s like ‘permission’ to continue.
Possibly a subconscious remark – possibly knowing (subconsciously) that the above comments did cause offence and perhaps the original poster (me) might not want to continue – so your subconscious tries to validate itself ‘please continue’ and just ignores what happened :palm:
OK .. I finished :popcorn:
Now we come to the most important part of this whole post / message to you.
Again I must clarify – all of my comments have good intentions behind them – there is no malice – I am doing this publicly rather via a PM because I don’t think there is anything here which is of a ‘private nature’ because everyone can benefit from this observation / comments – not just you.
1. You need to take time – read the post IN DETAIL – don’t skim through and just look out for the name ‘Siglent’, decide if there is an implied ‘invitation’ for other members to ‘chime-in’ – if not then don’t do it just for the sake of increasing your forum posts counter.
2. If there is good reason and an implied permission to ‘chime-in’ – do this with etiquette – don’t make tong-in-cheek-remarks (we all do it sometimes – but you need to refrain from this) and most importantly try to be helpful to the post in question , don’t simply make an unrelated remark – because this , believe it or not, will ‘offend’ (subconsciously) most of the time.
3. Please don’t wave your ‘distributor cap’ – this means don’t talk about any products , prices, features, new models, or ANYTHING ELSE relating to your business. Your only contribution to the post you respond to should be from ‘Rob’ the knowledgeable person and NOT ‘Rob’ the Siglent Distributor. THIS IS THE MOST IMPORTANT ‘TIP’ – because if you do this – its highly likely all of your small nit-picking offensive remarks will disappear in the process – because most of them are often related to ‘Rob’ the Siglent Distributor and most likely not ‘Rob’ the knowledgeable person.
4. If someone ASKS a specific question – to you directly – in the post – inviting your input, do this with care – ONLY answer the question!, don’t add additional information, don’t make judgements , don’t make side comments – unless specifically asked to do this.
I have tried to cover a very difficult subject in this post – the message is definitely up there in the ‘text’ – I tried my best (without writing a detailed essay - although this is a long post) – to highlight the most important aspects.
I used your reply to me – in THIS post as a great example (I am sure there are many others in the 18600 posts you made within this forum) – to highlight how a simple, what you think is a helpful reply, is in fact destructive and offensive instead.
I hope the above notes give you more insight to your – most likely subconscious - behaviour in the hope that it will HELP you make more meaningful posts & contributions to the forum in the immediate future.
Please once again don’t take offence Rob – we all love you tautech, BUT you need to be informed of your forum etiquette shortcomings – all for the benefit of the members here.
Take Care and Be Safe
-
Tautech, IMO modifying a post in a way to completely change its contents, especially if it had been referred to afterwards, is at least questionable. But I'm not the one to argue about that.
Yet, your modified contents is not exactly correct either:
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1028104)
There's a problem with this SSG3021X measurement.
First harmonic doesn't meet spec as it should be 30dB+ down.
SSG3021X 10 MHz 10dB using Siglent 6GHz N-N cable connected to SVA1032X in FFT mode with settings to reveal the 4th harmonic.
It's true, the absolute level of the second harmonic in @noreply's measurement is just shy of -30dBm, but the relevant figure is the damping of the harmonic vs. the carrier, which in the aforementioned measurement is at almost +13dBm. Hence, the second harmonic is close to -42.5dBc which is very well within spec. I'ld say the measurement that @noreply did is completely valid.
-
I tried to reproduce your recent harmonics plot – using same settings as shown on your screen capture – well at least the ones I can see as shown on the captured plot you posted.
First observation…
I tried hard to be able to have ‘same’ plot settings and at the same time for the markers to show ‘true’ dBm readings – specifically as shown on your plot.
Unfortunately I was unable to reproduce this.
Can you PLEASE advise – and possibly show (maybe screen capture) ALL of your settings?
When I applied the 10dBm ref – I was unable to get a ‘true’ direct dBm reading on the markers.
The 10dBm was ‘added’ to the measurements of the markers as seen on the screen capture I made.
I guess this is not going to alter the result – we simply subtract 10dBm from the marker readings to get a ‘true’ reading.
However I am puzzled at how you managed to get ‘true’ marker readings whilst still having the 10dBm ref set?
Once again please provide me with ALL the setting – so I can reproduce the harmonics response EXACTLY as you made – to have identical scaling settings – so our respective markers will have same initial conditions.
After all , we have the same HW – so this should be simple.
Second observation …
Once again, I was having problems in getting the SVA’s noise floor as low as you did.
I noted the EXACT same BW settings – can see this from my plot.
I also used the EXACT same frequency sweep settings – can see this on the plot.
And as a quick check the sweep time was EXACTLY same of 245.775 seconds – just like your plot.
So, can you please help me get in getting the extra -10dBm of noise floor?
I guess this has something to do with the 10dBm Ref – it ‘shifts’ the whole ‘plot’ by 10dBm.
Once again – it would be nice if I can share you exact setup – so can fully reproduce SAME conditions for the Harmonics test.
Third observation …
My frequency on the SSG was set to 10MHz.
The ‘true’ frequency from the SSG was actually 10.95769 MHz – despite this my SVA did not show ‘true’ frequency of the SSG.
The SVA is not capable (possible FW bug) – or it simply is unable to do accurate math's on a wide span when using markers?
This is a known behaviour – not just happening here.
However – looking at your screen capture plots – I can see you have ‘exact’ 10MHz clock.
This would be an exceptional achievement – given the problem I and possibly others have experienced with accurate frequency readings on the SVA with wide frequency span when using markers.
But – a closer look at your plot reveals an external REF – so I presume you ‘locked’ the SSG to the SVA or just provided an external 10MHz reference to both instruments?
I am not an expert on the internal signal processing of the SSG and SVA and the possible effects on harmonics – when ‘external’ frequency sources are used?
I guess for the sake of 'like–for–like' – maybe external references should not be used?
Final observation …
The harmonic frequency response I posted for the Rigol DSG815 was at its calibration limit – that of 13dBm (same level and limit as the SSG)
The plot you just posted – to compare the Siglent SSG3021X harmonic response was NOT to a 13dBm level – it was plotted to a 10dBm level.
As you know – even using your own expression – this is not ‘apples-to-apples’ is it not?
So I took the liberty to plot BOTH the 10dBm and the 13dBM responses – they are attached below.
I look forward to your screen shot instructions as to the ‘exact’ same settings as discussed above – so I can reproduce your noise floor and the ‘true’ marker readings and re-run the harmonic responses again.
Thanks tautech
-
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1028104)
There's a problem with this SSG3021X measurement.
First harmonic doesn't meet spec as it should be 30dB+ down.
SSG3021X 10 MHz 10dB using Siglent 6GHz N-N cable connected to SVA1032X in FFT mode with settings to reveal the 4th harmonic.
I can not see any problem in @noreply test image.
10MHz carrier aka fundamental aka 1st harmonic is in its level is displayed ok. Second harmonic, 20Mhz is well inside specifications around -42.5dBc
We can not see more harmonics due to used very wide (300kHz) RBW so 3rd and higher harmonics are under displayed noise level.
Used "Mixer level", btw, is in this @noreply image -37dBm, so SA own generated harmonics are nearly around as low as it can be. SA specs are given for -30dBm 1st mixer "mixer level". Att 50dB and signal level in input +13dBm so -37dBm.
It must say that Siglent signal generator is RF generator.
Rigol here is dds function generator. Even when they go also to same frequencies they are very different.
ETA: OMG.. somehow I think there was R's function gen. Perhaps due to used low freq and also these R's model numbers... DG and DSG... and yes @TT leas I did not note N connector... also perhaps R's overall size was so small... so my apologize messing things.. It DSG800 IS of course RF generator. Period.
Not for this case... I have seen many times, also Siglent function generators give less harmonics levels than example high quality HP RF generator, I remember I have made some this kind of tests example with some HP8642B. These old tests can see somewhere deep inside this forum if I remember right and also in many data sheets. Typically RF generators output power amplifier and after then attenuator and leveling stage is very different (with reason) than in typical function generators.
Take very narrow RBW and go to very close carrier and down to bottom...
-
If I understand @noreply's tests correctly, he has done the measurements on both generators at the N connector, so there's no DDS or whatever function generator business going on in case of the Rigol instrument. Since the photo function of the forum is still broken, it's only superficially visible in the thumbnails which connectors had been used.
So it is obvious that Rigol's signal generator has a much lower second harmonic vs. Siglent's counterpart. My reverse engineering of this device also indicates that the output stage is designed for way more power output than it will ever "experience" in this generator, so it's possible that it's pretty linear at the powers it's used at (provided it's a class A design with sufficiently high quiescent current).
Another detail that may be quite interesting are the "dimples" in the spectrum at 5MHz intervals in the test of Rigol's generator. I'ld love to see the same spectrum taken at a much lower resolution bandwidth (in order to lower the noise floor) and maybe some trace averaging. Is there really a "forest" of peaks somewhere at -50dBc? If there is, it may be interesting to check the higher bands, i.e. at a frequency that isn't generated by mixing (> 227.5MHz). Siglent's generator doesn't appear to show these artifacts.
-
Another detail that may be quite interesting are the "dimples" in the spectrum at 5MHz intervals in the test of Rigol's generator. I'ld love to see the same spectrum taken at a much lower resolution bandwidth (in order to lower the noise floor) and maybe some trace averaging. Is there really a "forest" of peaks somewhere at -50dBc? If there is, it may be interesting to check the higher bands, i.e. at a frequency that isn't generated by mixing (> 227.5MHz). Siglent's generator doesn't appear to show these artifacts.
@TurboTom
I'm waiting to get some feedback from toutech on his analysis settings - as questioned in my post?
If I do , I will repeat my test with the EXACT settings he used.
After reading rf-loop's input as well as your comments - I will do some specific 'low noise floor' narrow RBW to see any harmonic artifacts on the Rigol.
My guess is that there will be quite a few - but Rigol did a good job at suppressing them low enough that it makes no difference to the user - since its well below the published specifications.
rf-loop could be correct - the Siglent may have fewer harmonics - but hey must have had some difficulty in trying to suppress them below a significant 'specification' level.
I will run this again on the Rigol & Siglent - but with different SVA settings that toutech used - so we can get more details at the lower levels.
rf-loop implied that the DSG800 series is not a true RF generator (digitally synthesized) - this as you said goes against your findings and I am not qualified to this level where my input would have any bearings.
BUT
I am happy to make any test you suggest to verify (via analysis) and possibly give you and rf-loop some insights to determine the most likely operation and architecture of the DSG800 series.
Looking forward to your input.
Take Care
-
When unexplained measurement levels are made, due to the many and deep menus it is best to press Preset to return SSA/SVA to know settings and try again.
Previously SSG and SVA clocks were linked using the SSG 10 MHz OUT for frequencies to be aligned.
Not so for the following screenshot where both instruments are running independent of any reference.
This time @ 1 GHz 0dB and again looking for harmonics of which the first is again ~50dB down like it was for 10 MHz @ 10dB.
For such a wide sweep when it is already ~3ks/sweep further reduction of the BW to lower noise floor would only delay what is already an acceptable result.
Improved SSG3021X ;)
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1031022)
-
Thank you for your input tautech
I’m sure everyone appreciates the opportunity to ‘see’ more response plots – however rather than make measurements to verify if the Siglent SSG is indeed performing within its published specifications (as you pointed out) was not the objective of this thread.
What best serves the members of this forum and any other ‘viewers’ who have not registered – it to understand the behaviour and performance or EACH of the respective manufacturers products – namely their - RF signal generator offerings.
Weather Rigol or Siglent meets their publicized specification is not under investigation here.
What’s nice to see – from testing these devices – side-by-side – are any ‘clues’ in how the signals are generated, any benefits one approach might have against the other, and more importantly – identify any limitations or other design aspects which could be taken advantage during day-to-day use of the instrument.
So, I hope you understand, that simply taking measurements – without any direct ‘reference’ to the other product under test – serves little purpose in the scope of this thread.
Oh, just for the record, when testing ‘specifications’ – to see if Siglent or Rigol is indeed performing within their published harmonic levels, it is BEST to use the HIGHEST ‘in calibration’ output level of the device – which happens to be identical at +13dBm for BOTH of the instruments.
Performing a harmonics test at 0dBm – although useful data – is not a true indication that the instrument will have a similar performance at +13dBm and be within its published specifications at this output level as well – if that was the objective of the test.
OK, with the above out of the way, I can now report some interesting findings – at least for turbotom and rf-loop, who might be able to interpret these with relation to the respective devices architecture and inherent performance differences.
One notable screen capture which I would like to comment on is the – image 1 – “rigol 13dbm 200mhz harmonics.png”
Looking at this – we can see the significant ‘roll-off’ in noise floor after the fundamental frequency.
I don’t know what’s going on here – but if this is ‘by design’ – it’s pretty clever – because they just dropped the noise floor (and subsequent harmonics) by about 8dBm – VERY CLEVER.
I did EXACTLY same plot with Siglent SSG by just switching the respective outputs of the RF signal generators – this time to the SSG’s output N connector.
Please see – image 2 – “siglent 13dbm 200mhz harmonics.png”
You will note that the general response looks similar – but has a higher noise floor – almost exact same level – approx. 8dBm (visual observation from plot – unfortunately I did not measure at the time) that was somehow ‘reduced’ on the Rigol.
From this observation alone – it can be seen that Rigol is definitely doing something “in addition” to reduce the harmonic noise floor – and succeeded – otherwise the response(s) look very similar.
@turbotom, @rf-loop
If this is the ‘smoking gun’ that you and possibly rf-loop were looking to find – perhaps both of you can discuss?
I have made some additional plots and posting below – these are ALL at +13dBm – to stress the harmonic response – at various frequencies , and more importantly SVA’s frequency span.
I found that marker measurements can vary – depending on the frequency span under analysis – so was mindful of this by taking a range of readings and screen captures.
Hope this data is useful – looking forward to your comments / analysis.
-
After making above post - the 'already posted' images of previous posts got messed up
This IS A KNOWN PROBLEM - in the forum and Dave or moderators have not rectified.
Any present / past and future readers of this thread - please be aware of the above!
I don't know if this will be 'fixed' and when it is - weather the images will get corrected automatically :-\
-
One notable screen capture which I would like to comment on is the – image 1 – “rigol 13dbm 200mhz harmonics.png”
Looking at this – we can see the significant ‘roll-off’ in noise floor after the fundamental frequency.
I don’t know what’s going on here – but if this is ‘by design’ – it’s pretty clever – because they just dropped the noise floor (and subsequent harmonics) by about 8dBm – VERY CLEVER.
I did EXACTLY same plot with Siglent SSG by just switching the respective outputs of the RF signal generators – this time to the SSG’s output N connector.
Please see – image 2 – “siglent 13dbm 200mhz harmonics.png”
You will note that the general response looks similar – but has a higher noise floor – almost exact same level – approx. 8dBm (visual observation from plot – unfortunately I did not measure at the time) that was somehow ‘reduced’ on the Rigol.
Due to this forum image attachment total mess (how long it take to repair)
I need ask if these are now right
I do not know if yoyr words there is some kind of sarcasm or what ever but In my eyes it looks that Rigol rise noise level near carrier this 8dB and more far from carrier it drops to "normal" like it is all time in Siglent. So what is rising around carrier noise level in Rigol so weird looking shape. But is it so that these images are ok and not swapped or what ever mess...
This is why I ask dive more deep inside near carrier bottom corner. You have there more narrow RBW for use when look near carrier USB or LSB bottom corner. So it can perhaps show what is there or if it is some kind of bit weird PN.
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1031388;image)
named as Rigol
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1031392;image)
named as Siglent
Can you take both these carriers alone with lot of more narrow RBW so we can see what is there. This 1st image near carrier bottom looks extremely weird. So it is nice if you can dive more deep inside this..
But then need also give some note. Your signal level is around +13dBm and SSA attenuator is 20dB !!
And you are looking harmonics. Which one you want look, SA's os DUT's.
You run SA with -7dBm Mixer level! For analyze DUT's harmonics it is lot of too high level! Even when this is same for both signal generators but with this setup there can easy drop to trap. Because... I think no need explain.. So least it need cross check with other settings so that what ever it makes to noise floor but look how much different settings affect carrier and harmonics ratio.
-
Thank you for your insight rf-loop
I will get a 'zoom-in' for the area of interest.
Yes, you are right - near carrier the noise level is higher
I used an inappropriate reference to noise floor - whereas what I meant to signify is that the resultant harmonic levels dropped by this same amount.
My thinking (not necessarily correct at all) - was that the Rigol design - somehow 'dropped' the noise floor - because its the harmonics are certainly lower at higher frequencies.
Anyway I will run some analysis & post results ASAP
-
It's likely the close to carrier noise of the Rigol is due to some waveform type dithering.
Best,
-
@rf-loop
Here are a few new 'zoomed-in' plots
I had to offset the scales - so could 'fit' the area of interest - so please take the offset dBm into account when looking at the marker readings.
I then did a 'difference' on two randomly placed markers (6 and 7) - just so we can get the value of the 'drop' of the floor just after the fundamental frequency.
I started the plot at 10Mhz and went just after the 1st harmonic
Did another plot from 5Mhz to just after the fundamental of 200MHz
If you need any further testing of specific areas OR be 'zoomed' - just ask.
Hope this is useful :)
-
As @rf-loop already pointed out, the raised noise floor of Rigol's machine between DC and approx. 300MHz is peculiar. I relate this to Rigol's approach (https://www.eevblog.com/forum/testgear/upcoming-rigol-dsg815830/?action=dlattach;attach=1020928) to generate the frequency band of 9kHz to 227.5MHz by mixing the 910MHz of the reference with the LO frequency range of 682.5~901MHz. This mixer is followed by an amplifier and a low pass filter, all of these located on the "wrong" side of the PCB. It may well be possible that these components pick up / generate some noise and the low pass filter limits it right to the range that can be observed. The following test may be useful to understand the situation: keep the SA setting as in the initial test and increase DSG frequency step-by-step above 227.5MHz so the DSG switches ranges. If in this moment, the noise band vanishes, it's related to the band1 mixer/amplifier/filter circuitry.
One more peculiar thing I observed: Peak no. 7 @ 138.2MHz with approx. -56dBm is more or less identically present in both the spectra -- which makes me assume that this is some noise picked up by the setup elsewhere and not generated by the signal generators. You may want to test the "output signal" of the generators while powered down to verify this.
-
@turbotom
I did some more plots as you suggested - thank you.
Looks like you 'hunch' panned out :-+
The plot - see image below - tries to show the following;-
I recreated 'zoom-in' area of interest - and switched on some markers for reference - remember to take into account the large offset.
I then tried to make multiple traces - so we can easily see any differences.
Unfortunately on the SVA there are only 4 traces - also I discovered a bug - the text color will not match the trace color when you save the plot INVERTED file - so be aware of this.
So to avoid confusion with trace labels and mismatch in colors - I used a non inverted (black background) plot which is attached.
I checked the DSG - before starting traces - and did a full spectrum 'zoomed' sweep - there is NOTHING coming out of the DSG in its stand-by mode - in essence when the RF out is OFF - its OFF!
I then used same output level for ALL traces +13dBm - highest 'in calibration' setting for the Rigol (same as Siglent) and started traces at 200MHz
Second trace was at 210, third at 220, then I did two traces - not captured - just for me to check in fine detail - 227 - (was same as 220 - the DSG obviously did not switch ranges) , then at 228 - BINGO - DSG switched ranges!
I then did a captured trace at 230
So all the traces can be seen on ONE plot
200MHz, 210MHz, 220MHz and the 230MHz - where the DSG obviously switched at 227.5MHz as you suggested :popcorn:
OK, now that's this is clarified - despite the noise floor being similar to Siglent SSG - Rigol still has significantly lower Harmonic response - any comments / explanations for this observation - with relation to architecture / circuit design?
Thanks
-
A screenshot saved from a previous check of SSG3021X PN @ 10 MHz 10dB at Span's of 2 MHz and 2 KHz:
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1031964)
-
@rf-loop
Here are a few new 'zoomed-in' plots
I had to offset the scales - so could 'fit' the area of interest - so please take the offset dBm into account when looking at the marker readings.
I then did a 'difference' on two randomly placed markers (6 and 7) - just so we can get the value of the 'drop' of the floor just after the fundamental frequency.
I started the plot at 10Mhz and went just after the 1st harmonic
Did another plot from 5Mhz to just after the fundamental of 200MHz
If you need any further testing of specific areas OR be 'zoomed' - just ask.
Hope this is useful :)
1st question. I hope more words do not affect for more mess.
Image names tell 13dBm signal level. Is it true. Is it SA input level or externally attenuated generator output level.
If it is SA input level +13dBm it is far too high. Perhaps not make so much bad in this case but we can not be sure. So please use -30dBm mixer level to be sure SA works inside specs with its own generated harmonics etc products in first mixer.
Then please drop Generator level to example to -70dBm (200MHz carrier) and turn AM modulation ON, use 50Hz modulation and modulation depth 0.1% - if can not more low. Set SA Attenuator to 0dB and Span 1kHz or less and RBW 1Hz. As I told earlier I like to see near carrier and its bottom corner. It depends now lot of how this Rigol's extra noise widely around carrier goes and what it is.
Then look also how this wideband noise floor around 200MHz drops and how also these non harmonics spurs drop if they drop when signal is more low and you can also go more down with SA what have also its own phase noise what is next thing where you hit your head very easy when you are near carrier where it is more low than example 10kHz away from carrier what is not anymore near carrier.
Harmonics are not RF generators achilles heel. non harmonic spurs are and phase noise, residual FM etc things are.
Of course these are not at all high class RF generators, as can also see in specs and there is harmonics levels are not main thing at all. Look example some "state of art" RF generators +35dBc harmonics specifications and think why. Yes they can suppress these but why because they, pure harmonics, are not key factors in state of art RF generators.
-
@rf-loop
Thank you for your feedback.
Yes long 'text' description of images can be confusing - I will try to be more structured in the future.
So to clarify some of your questions and comments;-
The 13dBm I refer to is ALWAYS the output level of the DSG815 RF output (N connector).
I chose this level because its the highest 'in calibration' level and it would be a good indicator of the 'worst case' harmonic response.
I can totally see your comment about harmonics as not being the RF generators Achilles heel - especially if they are 'clean' and well behaved (indication of a good and pure sinusoidal source) - so like you said - spurs and phase noise can be more critical aspects of performance.
Ok, I have re-run the test plots with the DSG815 output settings you suggested.
I have set the SVA to operate at its lowest noise floor for the given parameters - including switching 'on' the preamp - to even further lower the noise floor.
RBW and VBW both set at 1Hz
On the DSG815 the frequency was set at 200MHz
Signal Level at -70dBm for plots marked, and -80dBm for plots marked as well as -65dBm for plots marked.
Modulation AM - switched ON with 50Hz and modulation depth at 0.1%
On the resulting plot(s) you can see a 'spur' at marker 2
I did a few additional plots - to see variation - as we are so close to the noise floor and resolution limit - so each plot looks a little different due to noise and resolution limit.
I then did a wider span (still at -80dBm DSG815 output level) - to cover at least the 1st harmonic.
From the resultant plot - there was no noticeable harmonics as can be seen on the plot.
So I increased the DSG815 signal to -65dBm and did a plot - where the 1st harmonic was visible.
Hope the above additional plots prove useful.
Thanks rf-loop
-
As promised before, I did a similar test with my newly acquired and repaired R&S (no, not Rigol&Siglent... ;)) SM300 RF Signal generator, albeit at 100MHz instead of 200MHz (shouldn't make much of a difference in case of a full-range mixer approach) and here, the amplitude modulation (100MHz carrier @ -70dBm, amplitude modulation 0.1% @ 50Hz) results in both sidebands properly being present... I hope the forum software won't mess up attachment as it did many times before... whatsoever, curious if
Rigol's Siglent's competitor will also only display a single sideband under these conditions...
Edit: P.S. It seems that the side bands at 22Hz offset are present on the SM300 as soon as modulation is enabled. Very strange... Noticed almost at the same time I pressed "Save" that I confused Rigol with Siglent :palm:
P.P.S. In this measurement, I synced the SSA3000X to the 10MHz output of the SM300 since there is some noticable offset between the two reference oscillators. Next time I'll pull out my Rubidium Standard (RS??? :palm: ;))....
P.P.P.P.P.P.P.S. Did some more tests with the SM300 to exactly replicate the previous measurements and at 3GHz, yet with the SSA3000X as analyzer. In how far this affects the results vs. the SVA1032X (plus) is something I cannot comment on.
-
whatsoever, curious if Siglent's competitor will also only display a single sideband under these conditions...
@turbotom
Just to be clear the investigations with rf-loop were on the Rigol DSG815 - this IS Siglent's competitor.
I am yet to do SAME test on the Siglent.
I will use same references on the Sig Gen (100MHz) as well as the SVA like you did - this way we should have 3 devices to compare (the R&S included)
Hopefully post results tomorrow ;)
-
@rf-loop
Thank you for your feedback.
Yes long 'text' description of images can be confusing - I will try to be more structured in the future.
So to clarify some of your questions and comments;-
The 13dBm I refer to is ALWAYS the output level of the DSG815 RF output (N connector).
I chose this level because its the highest 'in calibration' level and it would be a good indicator of the 'worst case' harmonic response.
I can totally see your comment about harmonics as not being the RF generators Achilles heel - especially if they are 'clean' and well behaved (indication of a good and pure sinusoidal source) - so like you said - spurs and phase noise can be more critical aspects of performance.
Ok, I have re-run the test plots with the DSG815 output settings you suggested.
I have set the SVA to operate at its lowest noise floor for the given parameters - including switching 'on' the preamp - to even further lower the noise floor.
RBW and VBW both set at 1Hz
On the DSG815 the frequency was set at 200MHz
Signal Level at -70dBm for plots marked, and -80dBm for plots marked as well as -65dBm for plots marked.
Modulation AM - switched ON with 50Hz and modulation depth at 0.1%
On the resulting plot(s) you can see a 'spur' at marker 2
I did a few additional plots - to see variation - as we are so close to the noise floor and resolution limit - so each plot looks a little different due to noise and resolution limit.
I then did a wider span (still at -80dBm DSG815 output level) - to cover at least the 1st harmonic.
From the resultant plot - there was no noticeable harmonics as can be seen on the plot.
So I increased the DSG815 signal to -65dBm and did a plot - where the 1st harmonic was visible.
Hope the above additional plots prove useful.
Thanks rf-loop
Tnx.
Some comment.
You have selected 100Hz Span and you use 50Hz AM modulation what produce carrier - 50Hz (LSB) and carrier + 50Hz USB) sidebands. So if yout carrier is center your LSB and USB sides are just display left and right border. 200Hz (or 250Hz like @TT images later). Siglent most narrow one FFT slice is 300Hz (with this setting) what is most small frequency hop when it is doing freq. hopping sweep in FFT mode. So with 300Hz Span or more narrow use same time and if even bit over 300Hz span then it use same time for 301 - 600Hz Span. and so on. (in some point, I do not remember these hops are more wide and if I remember right there is least 3 "gears" how it is hopping. so or so it do not affect here so much but only for sweep time and some cosmetics some times just in these slices borders because there is not any ideal things like ideal filters etc, these exist only in students books)
But, now... why there is not at all visible these sidebands in your images but @TT images have and very clearly. These need be even partially visible because your carrier is mostly not center and even if center half of sidebands shape need be visible due to 1Hz filter what have roughly 1:5 shape and 1Hz width is -3dB width.
Sidenote to some random readers. 10% AM mod depth (here with 50Hz modulating sinewave) need give sidebands level -26dBc and 1% mod depth -46dBc and 0.1% naturally -66dBc. Of course generators modulation depth accuracy is questionable. If 0.1% is 0.2% it is 6dB difference in AM mod sidebands. (double % equals 6dB)
If look your noise floor and carrier 0.1% AM mod sidebands need be still easy detectable if all is ok. Now all is not ok.
Also it can see that Rigol huge noise floor around carrier (previous 200MHz images) is dropped also down as carrier level is dropped. Is it some form of PN - without deep investigations can not know. Guessing is form of entertainment art so... let it be.
So, where are these 50Hz AM mod sidebands? Did you know why I ask -70dBm carrier :)
look friend @TT (TurboTom) images... where is noise level.
Make (no need image) small check... input example 100 or 200MHz from Generator A and generator B
Select 3Hz RBW (I know some reasons, specially if you use preamp) Set SA 0dB Att and quite narrow Span, example 3kHz and and ref line so that noise floor is near bottom and normal default detector as is ( and If I do this I do not use preamp). Set carrier near other side of display (because it is not important which one sideband you look and you can use more narrow Span for reduce waiting time).
Then start watching noise floor when you rise generator output starting from generator bottom, example -143dBm (oh but these gens do not go so low so start anyhow generator bottom level) continue rising until generator out is ref level.
Look if both A and B generator give same result.
As can see @TT images noise floor near 100 and 200MHz carrier with signal levels -70 and -80dBm
compare with your images here.
For all testers. Even if you have super good or what ever external freq standard. These tests are best to do without because all do not have and more importantly, no one know what these different ext references do to Siglent SA phasenoise independent of ext ref own phase noise. This whole case is extremely unclear. So for avoid unknown to come disturb.... freq accuracy is not question here at all. This is totally other thing. So why take risk and keep door open for unknown until there is enough evidence that ext ref really can not disrturb Siglent SA phase noise even if ext ref own PN is low. .
Try to find how you can find AM mod .1% sidebands near carrier. Or is reason in generator what have so dirty near carrier. ;)
-
As promised before, I did a similar test with my newly acquired and repaired R&S (no, not Rigol&Siglent... ;)) SM300 RF Signal generator, albeit at 100MHz instead of 200MHz (shouldn't make much of a difference in case of a full-range mixer approach) and here, the amplitude modulation (100MHz carrier @ -70dBm, amplitude modulation 0.1% @ 50Hz) results in both sidebands properly being present... I hope the forum software won't mess up attachment as it did many times before... whatsoever, curious if Rigol's Siglent's competitor will also only display a single sideband under these conditions...
Edit: P.S. It seems that the side bands at 22Hz offset are present on the SM300 as soon as modulation is enabled. Very strange... Noticed almost at the same time I pressed "Save" that I confused Rigol with Siglent :palm:
P.P.S. In this measurement, I synced the SSA3000X to the 10MHz output of the SM300 since there is some noticable offset between the two reference oscillators. Next time I'll pull out my Rubidium Standard (RS??? :palm: ;))....
P.P.P.P.P.P.P.S. Did some more tests with the SM300 to exactly replicate the previous measurements and at 3GHz, yet with the SSA3000X as analyzer. In how far this affects the results vs. the SVA1032X (plus) is something I cannot comment on.
It is good there is some third brand "reference" like R&S...
(what have specified as: SSB phase noise <-95 dBc (fc = 1 GHz, 20 kHz offset, 1 Hz measurement bandwidth)
<> Siglent 3000X Phase Noise : -110 dBc/ Hz @ 1 GHz , 20 kHz offset (typ.)
<> Rigol DSG800 Phase Noise : 100 kHz ≤ f ≤ 1.5 GHz < offset = 20 kHz, -105 dBc/Hz, < -112 dBc/Hz (typ.))
.... and so that different tests here by forum members have made "enough equally" and even so that also tested with same kind of equipment (Sig SSA even when they may have small differences. One lab - all equal is not possible here)
If I'm in homeland home I can also do some tests but... now impossible. I can here only play with some bit better dual beam IR CO2 sensors and crap single beam ones and some poor and some bit better laser scatter small particle detectors and so on... for make some environmental datalogger... when peoples come old it can see that time is really proportional(//)... more age and more fast it goes. I do not know if time is speeding up or I'm going more slowly... or is it so that I try do more things than when I was younger.
-
@ rf-loop
My sincere apologies for the rather useless plots I made.
You must remember - although I understand basic RF principles - compared to you and turbotom I am a total 'noob' and this with the combination of my poor understanding of your instructions - I totally 'missed the boat' with my plots - as I completely missed the area of interest which should have been captured.
It was not until turbotom posted his plots / results for the R&S SM300 - that I realized my mistakes |O
Thank you for your explanation rf-loop, I appreciate your efforts especially when English is not your mother tongue :clap:
To avoid any further confusion, I have made plots BOTH for the Rigol DSG815 and Siglent SSG3021X using identical parameters as turbotom did in his post above - Reply #25
Now we should have all three different manufactures for a quick comparison.
The respective plots are attached below with references to the turbotom's plots which they should mimic.
Hope that these are now sufficient - for some sensible commentary and discussion on the obtained results
Thanks @rf-loop, @turbotom :)
-
I have been trying to do some simple EXT Modulation testing with the SSG today.
Unfortunately, I'm not having much luck - and getting some strange observations.
This is what I am trying to do;-
I have a stereo encoder module.
The module simply creates a 19KHz FM stereo 'pilot' - and has LEFT and RIGHT audio inputs.
When 'powered-up' and output connected to SVA - you can clearly see the 19KHz pilot.
I then - on the SSG - want to set a Frequency of 108MHz with a level of 0dBm - this is no problem - and when 'set' and the RF output enabled on the SSG - I can clearly see the 108MHz carrier on the SVA at pretty much the 0dBm level.
Now I would like to connect the stereo encoders 'output' to the EXT Modulation INPUT (BNC) of the SSG.
This should modulate a 19KHz 'pilot' - the FM stereo pilot onto the 108MHz carrier.
Despite selecting the FM external modulation menu on the SSG and enabling the output - I cannot 'see' the 19KHz 'pilot' - it should be exactly 19KHz offset from the 108MHz carrier.
What am I doing wrong?
I presume that there is something wrong with my process :-\
I performed a similar test on the Rigol DSG815 and there was no problems - could clearly see the 19KHz 'pilot' next to the 108MHz carrier.
I even then got my portable FM radio - tuned to 108MHz and put close to the N connector input on the SVA - there was enough signal for me to hear the modulation test on the radio - in stereo.
My input to the stereo encoder module was audio from a YT video doing LEFT & RIGHT speaker testing - which was faithfully reproduced on the FM radio with the Rigo DSG815 acting as a FM Stereo TX.
So the proof of concept works on the Rigol DSG815 - but I cannot do same on the SSG :(
Any suggestions in configuring the SSG would be most welcome :)
-
The external modulation input on the Siglent has an impedance of 50 ohms and is probably loading your source (the stereo encoder).
The Rigol has a selectable impedance and probably defaults to high impedance so it does not load down your source.
This is an example of why it's so hard to choose a complex product like a signal generator and why it's important to have detailed specifications from the manufacturer.
-
I got bitten by this issue myself.
I had a certain test that I usually ran using an Anritsu MG3642A generator. It always worked fine.
One day I had to run the test and I had an Agilent ESG on the bench so I used that. The test didn't work.
I learned the the external modulation input on the Anritsu is 600 ohms - which makes sense for an audio input, whereas the ESG has a 50 ohm input - something I would never have suspected.
So I checked and my R&S SML has a high Z input. WTF? Apparently, there isn't a standard.
Props to Rigol for giving you a choice of all three so their unit will work with any test setup you have been using.
-
@xmo
Thank you for your input.
Guess what?
You are 'spot-on' with your comments / observations :-+
Who would have thought that Siglent - with all their wisdom - in the extremely 'modern' SSG3021X (compared to the 2015 vintage DSG800 series) would have made such a critical omission in their design - by not allowing the ability to 'set' the impedance of the external modulation input |O
From this single omission - they just rendered the EXT input modulation capability pretty much useless for anyone wanting to use external modulation from professional audio and RF equipment.
I did notice the full capability of the Rigol DSG800 series in allowing the EXT modulation input impedance to be varied comprehensively - and it defaults to 600ohms, so I did not consciously 'look' for this in the Siglent SSG - but even if I did - its nowhere to be found.
I will ALWAYS stand to be corrected - because it is possible to be blinded by all the flashy menu options - and perhaps there is a secret menu - where indeed you can set EXT Modulation Input impedance ::)
Well, for a guy who is willing to spend over $2K on a RF signal generator and is highly dependent on EXT modulation - for testing RF circuits (remember its sometimes easy to generate a complex modulation with an external source - like an SDR or a custom HW module like a stereo encoder - I am yet to test the RDS encoder) - the Siglent has just become a non contender :palm:
I must admit - I like the modern interface AKA touch screen - but if the device is unable to do the 'job' interface is not important.
I hope that since we now have some noise floor measurements - someone can chime-in on how the Rigol performs against the Siglent?
A quick observation reveals Rigol has slightly better figures :-[
Finally, I hope that the EXT Modulation Input impedance is a 'forgotten' feature only in FW and does not need a HW revision to add the additional termination circuitry :popcorn:
-
I have been doing a little bit of |O because I was hoping that Siglent SSG3000X would be able to have 'high impedance' External Modulation Input.
So, I decided to have a look at the PUBLISHED SPECIFICATIONS for the SGS3000X series
- The DATASHEET
- The USER MANUAL
I Can confirm that in both of the above documents (SEE CAPTURED IMAGES) - Siglent confirms that there is only a 50 Ohm Input |O
So, I decided to move UP to the SSG5000X Range - maybe there is a high impedance EXT Modulation Input on this more recent and more expensive offering - perhaps Siglend realized the omission on the SSG3000X range and made sure this did not happen on the 5000X range ::)
So I decided to have a look at the PUBLISHED SPECIFICATIONS for the SGS5000X series
- The DATASHEET
- The USER MANUAL
BINGO!
The DATASHEET - sort of confirms that there IS a HIGH IMPEDANCE EXT Modulation Input (SEE CAPTURED IMAGES).
BUT
When looking at the USER MANUAL - IT clearly states that there is only a 50 Ohm EXT Modulation Input Impedance - NO SIGN of a 'high impedance' input as suggested in the DATASHEET |O
OK - so even after moving to the over $5K product range of the SSG5000X - there is still confusion as to weather Siglent offers the same flexibility of selecting various impedances for the EXT Modulation Input as on the Rigol DSG800 Series - a much lower cost product - first released in 2015 :popcorn:
Message to Siglent
PLEASE
FIX the EXT Modulation Input on the SSG3000X range to support 50 Ohm, 600 Ohm and High Impedance (>1Mohm) selections, via the user interface.
PLEASE
Advise if this can be a FW 'fix' - or weather it requires modifications to internal circuitry.
Thank You
-
OK. Here are a couple more thoughts for you about external modulation capability in signal generators.
Will the AM external modulation capability accept signals down to DC? Why do you care? If you can disable the generator's ALC and input DC to the external modulation input, you can create test configurations with external leveling.
Guess which generator goes down to DC and which one doesn't.
How about external FM modulation down to DC? Why do you care? If you can FM modulate down to DC you can create test setups where the generator can function as a VCO.
Same question: which generator goes down to DC and which one doesn't?
So, as rf-loop pointed out, there's a lot more to selecting a signal generator the harmonic performance.
It would be really nice if there was a single resource with a comprehensive comparison of multiple signal generator models, their functions, features, and specs.
Speaking of specs, one spec rf-loop mentioned is phase noise. I am really surprised that neither the Siglent or Rigol is really very good with the SSG at -110 and the DSG at -112.
[ that's dBc in a 1 Hz bandwidth at 20 kHz offset from a 1 GHZ carrier]
A look at some 20 year old generators:
Agilent ESG: -116 dBc
Agilent ESG-DP: -134 dBc
R&S SML: -128 dBc
Why do you care? For testing communications equipment you need a generator with low phase noise. Here's what ANSI/TIA-603 has to say about signal generator requirements:
"The RF signal generator shall have the following characteristics:
...
c) Single sideband phase noise in a 1 Hz bandwidth of 135 dB below the carrier at 25 kHz offset from the carrier for frequencies of 500 MHz and lower.
For frequencies above 500 MHz, the phase noise shall be less than130 dB below the carrier at a frequency offset of 25 kHz from the carrier."
So neither of the subject generators would be suitable.
BTW, you can't test the phase noise of either of these generators with the Siglent SSA because it's own phase noise is much higher than either generator.
For a tutorial on phase noise, see 'The Signal Path' #162
-
OK. Here are a couple more thoughts for you about external modulation capability in signal generators.
Will the AM external modulation capability accept signals down to DC? Why do you care? If you can disable the generator's ALC and input DC to the external modulation input, you can create test configurations with external leveling.
Guess which generator goes down to DC and which one doesn't.
How about external FM modulation down to DC? Why do you care? If you can FM modulate down to DC you can create test setups where the generator can function as a VCO.
Same question: which generator goes down to DC and which one doesn't?
So, as rf-loop pointed out, there's a lot more to selecting a signal generator the harmonic performance.
It would be really nice if there was a single resource with a comprehensive comparison of multiple signal generator models, their functions, features, and specs.
Speaking of specs, one spec rf-loop mentioned is phase noise. I am really surprised that neither the Siglent or Rigol is really very good with the SSG at -110 and the DSG at -112.
[ that's dBc in a 1 Hz bandwidth at 20 kHz offset from a 1 GHZ carrier]
A look at some 20 year old generators:
Agilent ESG: -116 dBc
Agilent ESG-DP: -134 dBc
R&S SML: -128 dBc
Why do you care? For testing communications equipment you need a generator with low phase noise. Here's what ANSI/TIA-603 has to say about signal generator requirements:
"The RF signal generator shall have the following characteristics:
...
c) Single sideband phase noise in a 1 Hz bandwidth of 135 dB below the carrier at 25 kHz offset from the carrier for frequencies of 500 MHz and lower.
For frequencies above 500 MHz, the phase noise shall be less than130 dB below the carrier at a frequency offset of 25 kHz from the carrier."
So neither of the subject generators would be suitable.
BTW, you can't test the phase noise of either of these generators with the Siglent SSA because it's own phase noise is much higher than either generator.
For a tutorial on phase noise, see 'The Signal Path' #162
Least my old work horse HP8642B have AC and DC FM and also Ext Mod impedance is 600 ohm.
Only "bit" more expensive (in its time) and heavy than these low level modern Sig and Rig. There is warning label... for 2 person carry... but classified for "silent room" radio tests with low RF leakage. And of course.. ancient but what I trust this I use. Simply.
Years ago I tell some Siglent person that why they do not take time and download all old HP Journalls and study and learn something about design philosophy. It is not so that young engineer sit behind table and start desing new lab equipment. First need have three things. experience, experience and experience + knowledge. Some can learn from others. no need copy ancient things but many times there is tens of years experience and high knowledge behind old instruments design and lot of these are told for everyone if can read. There is totally free materials for this. Electronics is different but test equipments basic usage have been so same tens of years that it is totally wood head who think external modulation input impedance is 50 ohm and period. Just because lack of enough experience and knowlege about anything from real world.
It must ask why these peoples are so lazy to find things and self learn - if can not ask. Yes I know partially one base reason it is in some country school system but it do not explain all... do NOT think... listen teacher.
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1034992;image)
-
I just did some testing on the Rigol DSG815 - slapped together a stereo and RDS encoder - now I have a full 'broadcast quality' modulated FM signal that can be pumped into a suitable RF generator with a properly conditioned EXT Modulation input.
The Rigol did the job - took some pics and posted below - I guess to show Siglent to get their act together :P
For about $100 worth of parts - you can now do FM receiver testing with Stereo and RDS - the flexibility of EXT Modulation Inputs should never be underestimated when buying a Sig Gen.
The Siglent SSG3021X has some very nice 'modern' features - but like rf-loop so eloquently put it - the 'young engineers' should take more care and LOOK CLOSELY at the OLD designs - there are countless hours of invaluable design experience which went into these products - so COPY the good features don't omit them.
The things I really liked about the Siglent SSG3021X - is the ability to set an 'offset' in the clock frequency - if you have a GSPDO - and can get atomic accuracy - its easy to calibrate the SSG within .01Hz
The other great feature is to be able to again set an 'offset' for signal level - another fantastic user feature - again if you have a more accurate reference - than you can 'dial' this up. Very handy when you have a lossy cable - and want to eliminate this from your signal loop calculations - just via this user menu :clap:
Anyway, its always good to be able to 'learn' from the old masters - so perhaps @rf-loop could make a final comment on the noise figures posted above - as to comparing the Rigol to the Siglen - they are pretty close - but is there a winner?
-
The Siglent SSG3021X has some very nice 'modern' features - but like rf-loop so eloquently put it - the 'young engineers' should take more care and LOOK CLOSELY at the OLD designs - there are countless hours of invaluable design experience which went into these products - so COPY the good features don't omit them.
The things I really liked about the Siglent SSG3021X - is the ability to set an 'offset' in the clock frequency - if you have a GSPDO - and can get atomic accuracy - its easy to calibrate the SSG within .01Hz
The other great feature is to be able to again set an 'offset' for signal level - another fantastic user feature - again if you have a more accurate reference - than you can 'dial' this up. Very handy when you have a lossy cable - and want to eliminate this from your signal loop calculations - just via this user menu :clap:
Anyway, its always good to be able to 'learn' from the old masters - so perhaps @rf-loop could make a final comment on the noise figures posted above - as to comparing the Rigol to the Siglen - they are pretty close - but is there a winner?
It is difficult with this EEVblog forum due to total mess with attached images. Example now I can not at all look these images where you make equal tests like @TT. Only small thumbnails are visible but when try look normal size image they are swapped with some other images. And original full size images are not in my cache so I can not see them anymore for further look.
But what I remember is that they are very close, some is tiny bit better in Rigol and some bit better in Siglent, if look these noise figures near carrier and also detected sidebands but perhaps there was one image what clearly show that Rigol AM mod level in 0.1% area was bit off. Overall Siglent low level AM mod levels looks bit better but also now only what I remember when I earlier look these.
Phase noise. SVA1032X is not at all right tool for this. First its own phase noise is problem for analyze these generators PN. This need Much better instrument. Also its PN is not fully specified. But if it is like SSA3000X then I know something about close carrier things and these can find also somehere in this forum, tested with my ex HP8644B.
We do not know (reading data sheet) enough about SVA1000X close carrier area PN until someone make some reliable tests (perhaps there also is but I have not looked). Also its specified PN is not ok for characterize these generators PN. In this test, related to phase noise we do not know if wagons are pushing or horses dragging. But there is some signs that these both are perhaps better than @TT 's R&S SM300 (if he have not disturbed its or SSA noise figure with some extref).
Amplitude accuracy. But +/- 1dB is normal good and typically it may be bit better.. Yes with offset can do correction if know used frequency and level true error but it can not make swept flatness better. And true accurate level measurement over freq scale and over level scale is not simply and easy. There come very easy things like are in time nuts, freq nuts and voltage nuts. Example many peoples have "religion" that GPSDO and all is ok. This is partially one big bull and shit. It can be good but it can also be total trap if blind believe. Take one state of art DOCXO free run and compare short time errors.
With SSG5000X user can do his own flatness correction data table - but 5000 is other case.
Here attached part screen what I see now there instead of test result images.
Who hell have made this total mess in Forum system related to attacments please go back to kinderkard. btw, years it have worked ok. Why someone have started repair well working system and result is this <censored>.
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1035530;image)
as can see original images are swapped with other images (compare @noreply original post)
Where this error is hiding?
So I recommend attach images inside msg and not only like thumbnail after msg IF it may help or not..
-
Who hell have made this total mess in Forum system related to attacments please go back to kinderkard. btw, years it have worked ok. Why someone have started repair well working system and result is this <censored>.
Maybe it's time to stop patching FW and starting patching the forum. :palm:
I have a feeling this all started with the new in-message images changes... :rant:
-
@rf-loop , @tv84
I have edited the posts with messed-up images and resubmitted the images again - they appear to be ok - for now :-\
Lets hope The Forum Admin staff are able to 'fix' this problem.
@rf-loop, perhaps you can now have a loser look at the plots if you wish to reexamine once again ;)
Thank You
-
@rf-loop , @tv84
I have edited the posts with messed-up images and resubmitted the images again - they appear to be ok - for now :-\
Lets hope The Forum Admin staff are able to 'fix' this problem.
@rf-loop, perhaps you can now have a loser look at the plots if you wish to reexamine once again ;)
Thank You
still
"Siglent SSG3021X - 1st plot - direct comparison to turbotom PNG3 plot.png"
give wrong image to me. It is some Siglent front panel picture. This is now only missing image what I do not have.
More fun when I try download it I get file but when I try open it it tell that it can not open because it is not png image. After I change filetype to jpg I can open it and then there is same front panel image what also can see in forum if click this small image.
-
More fun when I try download it I get file but when I try open it it tell that it can not open because it is not png image. After I change filetype to jpg I can open it and then there is same front panel image what also can see in forum if click this small image.
:palm: :palm: And no word from the admins...
-
@rf-loop , @tv84
I have edited the posts with messed-up images and resubmitted the images again - they appear to be ok - for now :-\
Lets hope The Forum Admin staff are able to 'fix' this problem.
@rf-loop, perhaps you can now have a loser look at the plots if you wish to reexamine once again ;)
Thank You
still
"Siglent SSG3021X - 1st plot - direct comparison to turbotom PNG3 plot.png"
give wrong image to me. It is some Siglent front panel picture. This is now only missing image what I do not have.
More fun when I try download it I get file but when I try open it it tell that it can not open because it is not png image. After I change filetype to jpg I can open it and then there is same front panel image what also can see in forum if click this small image.
The images seem to be OK for me - when I view now - after I reuploaded them.
Here are the Siglent images once again... hope they don't get messed-up
Thanks @rf-loop
-
tnx
finally get all images. Later...some comment but of course with just some single test images can not say lot.
Also it looks like there is ExtRef used so it make things more questionable because we do not know how it affect other things than average frequency. So there is one or more unknown included. It affect more or less, lets hope less. But from images effect can not see so I name it added unknown. I do not even understand why it is used. If we look non harmonic spurs, sub harmonics and harmonics, phase noise, modulation, levels etc there really do not need External Reference for correct some small sub ppm offset in RF generator. Then when we need very accurate freq and example so that all bab equipments use same reference, even if it is bit off or not then need ext ref. But, before use Ext Ref it is good to know how it MAY affect also other things than just freq average accuracy.
Oh and where is now Rigol DSG815 - 3rd plot - direct comparison to turbotom PNG8 plot.png (80.91 kB, 1023x600)
Name is there but image is Siglent 200MHz. (there is now 2x this same image in this same message where is Rigol and Siglent images together and there is now Rigol 1.5GHz missing when I look it now.)
-
Oh and where is now Rigol DSG815 - 3rd plot - direct comparison to turbotom PNG8 plot.png (80.91 kB, 1023x600)
Name is there but image is Siglent 200MHz. (there is now 2x this same image in this same message where is Rigol and Siglent images together and there is now Rigol 1.5GHz missing when I look it now.)
@rf-loop
I posted the 3 Rigol plots above - so should have both Siglent & Rigol now ;)
-
Today I decided to think 'outside the box' with regard to the Siglent SSG EXT Modulation Impedance being 'fixed' at 50 Ohm.
So how can we 'fix' this without breaking inside the SSG?
Simple solution is to build a 600 Ohm to 50 Ohm balun - this is a 12:1 ratio and can be build and tested - to provide a good impedance match.
If you are new to balun construction - here is a good guide ...
http://www.zl6qh.com/000225.html (http://www.zl6qh.com/000225.html)
- it describes an Antenna for RF transmission - hence large and bulky balun to handle the power - we don't need anything like this - so its only a guide in the design methodology - you can easily make your own 'small' mini balun for the use with the SSG.
It could even be mounted inside a 'box' with BNC In / Out connectors.
But the balun solution is a 'lossy' solution and may not be ideal for some type of low power modulated input signals.
Another approach is to have an active OP Amp where we can control the input impedance - make it high and also the output impedance - make it 50 Ohm.
Being Lazy - at least today (it was like 30 Deg Celsius in London Today) - so I decided to use a 'off the shelve' module - in essence an OP Amp - preamp for phono cartridge pick-up.
Here I simply used one channel - to feed my high impedance (600 Ohm) output from the RDS / Stereo Encoder - and the output of the OP Amp was close to 50 Ohm - so this was connected directly to the SSG's EXT Modulation Input BNC connector.
Guess What?
After setting up the 108MHz at 0dBM and applying a the RDS / Stereo encoded signal on the EXT modulation BNC (through the OP Amp) - switched on my little FM radio - and Voilà - Music to my ears - everything working as expected.
This is not perfect and the 'audio level' is quite low - but it is proof of concept - we can 'fix' the EXT Modulation Input Impedance to accept High Impedance and not only 50 Ohm.
The best way to do this is to have Siglen make a FACTORY Module - a little 'box' with ACTIVE circuit - to be able to accept a BNC input and then provide a 50 Ohm output impedance to the SSG's EXT Modulation Input BNC connector.
This 'box' could have a 'switch' that could have 3 settings
- High Impedance > 1Mohm
- 600 Ohm Impedance
- 50 Ohm Impedance
This WILL WORK - and Siglent could even have an adjustable 'level' control to increase the amplitude of the modulated signal - a bonus feature - because sometimes a 'raw' modulation signal can be extremely low level - such as an output from an SDR where you are creating the output in software - but the SDR had limited output power.
So what did we achieve here?
Made a little experiment - proof of concept - that it is possible to 'fix' the Siglent SSG30231X factory 'omission' of a selectable impedance on the EXT Modulation Input.
If such a FACTORY DESIGNED module can be made promptly & supplied with EVERY SSG - FREE OF CHARGE - then the SSG once again can become a respectable instrument for day to day use where an EXT Modulation High Impedance Input is mandatory requirement.
If anyone associated with Siglent FACTORY (yes this could be you tautech ;) ) is reading this post - then PLEASE feel free to copy this suggestion and hopefully in two shakes of a tail (however the old saying goes ...) - we will have a little module to plug into out SSG that will fix this problem!
-
@turbotom , @rf-loop and anyone else ...
I have been looking at some 'old' RF signal Generators and in particular their Specifications.
Its interesting to see how 'good' their specifications really are - considering their age :clap:
https://maximinstruments.com/pdf_files/MARCONI_2023_Datasheet.pdf (https://maximinstruments.com/pdf_files/MARCONI_2023_Datasheet.pdf)
https://accusrc.com/product-Rohde-Schwarz-SM300-8340 (https://accusrc.com/product-Rohde-Schwarz-SM300-8340)
https://www.testequipmentdepot.com/usedequipment/pdf/8642B.pdf (https://www.testequipmentdepot.com/usedequipment/pdf/8642B.pdf)
From a 'service' / calibration point of view the Marconi Instruments devices seem to be 'build-like-a tank' with a high degree of discrete components and a full BOM list and circuit diagrams in the user manuals - sorry did not see any teardown videos on the R&S SM300 or the HP 8642B as yet - but Dave did a good job at a Marconi Instruments 2025 (I hope I got the version correct - but they all share same structure) teardown.
So, why is it that some 20 years later - with all the advances in analog semiconductor technologies and modern manufacturing techniques - there still is no COMPETATIVE offerings(***) from the major manufactures - including China based - like Rigol and Siglen - to name just two of the most familiar here on this thread?
(***) Competitive offerings - in relation to a good condition 2nd hand specimen of one of the above listed, the price will often be BELOW the cost of an inferior specification 'new' device from all major manufacturers.
With this observation alone - I really don't see why anyone would buy a new RF Signal Generator - especially within the SUB $3000 range - when for this price alone you could score a 'old' high spec device which will beat anything on offer.
Any comments / discussion welcome
Yeah .. subject is a little 'off topic' but good to know the reason why we get what we get from the current manufactures within this price range
- does this mean that the current generation of products in this price range are OVERPRICED?
- if they were well 'below' the cost of a GOOD 2nd hand specimen of the 'older' generation
- would you still buy an 'oldie' or have the modern looking touch screen variant with 'peace of mind' that you don't need to fix-it yourself
- mind you if it still works after 20 years - it will work for another 3 (typically the new device warranty)
What do you think?
-
@turbotom , @rf-loop and anyone else ...
I have been looking at some 'old' RF signal Generators and in particular their Specifications.
Its interesting to see how 'good' their specifications really are - considering their age :clap:
https://maximinstruments.com/pdf_files/MARCONI_2023_Datasheet.pdf (https://maximinstruments.com/pdf_files/MARCONI_2023_Datasheet.pdf)
https://accusrc.com/product-Rohde-Schwarz-SM300-8340 (https://accusrc.com/product-Rohde-Schwarz-SM300-8340)
https://www.testequipmentdepot.com/usedequipment/pdf/8642B.pdf (https://www.testequipmentdepot.com/usedequipment/pdf/8642B.pdf)
From a 'service' / calibration point of view the Marconi Instruments devices seem to be 'build-like-a tank' with a high degree of discrete components and a full BOM list and circuit diagrams in the user manuals - sorry did not see any teardown videos on the R&S SM300 or the HP 8642B as yet - but Dave did a good job at a Marconi Instruments 2025 (I hope I got the version correct - but they all share same structure) teardown.
So, why is it that some 20 years later - with all the advances in analog semiconductor technologies and modern manufacturing techniques - there still is no COMPETATIVE offerings(***) from the major manufactures - including China based - like Rigol and Siglen - to name just two of the most familiar here on this thread?
(***) Competitive offerings - in relation to a good condition 2nd hand specimen of one of the above listed, the price will often be BELOW the cost of an inferior specification 'new' device from all major manufacturers.
With this observation alone - I really don't see why anyone would buy a new RF Signal Generator - especially within the SUB $3000 range - when for this price alone you could score a 'old' high spec device which will beat anything on offer.
Any comments / discussion welcome
Yeah .. subject is a little 'off topic' but good to know the reason why we get what we get from the current manufactures within this price range
- does this mean that the current generation of products in this price range are OVERPRICED?
- if they were well 'below' the cost of a GOOD 2nd hand specimen of the 'older' generation
- would you still buy an 'oldie' or have the modern looking touch screen variant with 'peace of mind' that you don't need to fix-it yourself
- mind you if it still works after 20 years - it will work for another 3 (typically the new device warranty)
What do you think?
I have still my old work horse HP8642B. It is quite good for normal conventional signals. Of course it can not do todays some important modulations and fast hopping etc. If you need QAM then game is ower with these oldies.
Why I use it. Because I trust it, it never fool me. Also it is overall in nice conditions. No need look more than this fron panel encoder. Bit different what these new generators have. Also if look these Rig and Sig. Put these to silent room and you can not do anything but close door and go home until they are cleaned away...
But it is heavy!
Download HP web archives "HP Journal 1985 December" you can read lot of about HP8642B design. Just for one tiny detail. Even LCD display they have thinked and keep care that it do not emit RF noise around. You can think these modern Rig and Sig and they TFT panels. Take antenna and listen these. Aften then I ask do you want these jammers to silent room. No, they are not even made for yhis kind of serious use.
And yes I have also its precessor, totally different HP8642, mechanically like just out from factory. Its frequency walk and drift here and there but it have amazing low phase noise also due to its cavity resonator.
Here: (There can also find lot of older HP journals and more new HP journals and some of them have lot of knowledge and basics for learning also nice develop and design culture)
https://hpl.hp.com/hpjournal/pdfs/IssuePDFs/1985-12.pdf (https://hpl.hp.com/hpjournal/pdfs/IssuePDFs/1985-12.pdf)
For conventional signals and analog type modulations etc old Good quality generators are still very good and far over these Rig and Sig new ones signal quality and they are made like tank starting from front panel encoder ball bearings in some models and of course they all are optical encoders not switch type cheap or cheapest junk (yes in world there is also good switch type rotary encoders example from Grayhill but also they are not cheap=. Just like also older state of art class spectrum analyzers (what Tektronix have never made).
But, how to buy for get good one. It depends many things. Least it need real knowledge about these.
Of course after then there come more modern but still old generators example from HP-destroyedHP aka Agilent-Keysight and of course from Marconi-IFR-Aeroflex.
Also some R&S. But all these companies have made generators for different purposes and everyone have made also quite poor models. Really old analog time RTV repair service did not need other than some total junk Heathshit or if have lot of money some R&S or HP bottom class model. Buyer need know what he is buying and mostly these better ones, originally very expensive State Of Art class (higher than High End class) generators are not cheap even now but quite cheap - depending how you think. many tell that they are old and unreliable. Today these modern generators are made for up to some years longevity in light use. Old times thinking was very very different. I have owned lot of, example HP and some R&S, gears what are not collected from waste loads and what have worked in heavy use 10 - 20 years and even still after then working in hobby like use lot of years and continuing. If also looks how complex these circuits are it is small miracle.
But back to original topic.
I have looked these some your test images. I leaved out these @TurboTom images due to reasons. First it is not Rigol Siglent compare. But nice reference here also other ways. One reason is that it goes to 3GHz and this image is not useful to compare 1.5GHz examples. Also it looks like there is something wrong in this individual test setup or generator. Last reason is also that if look data sheet it tell that this gen is in different more poor class with Phase Noise than Rig and Sig what are very close each others. Rigol datasheet tell it is tiny bit better but these images do not tell it or make this claim even questionable (image mark C).
What can see there is that Rigol AM modulation level is "out of order" but I have not checked if they have promised anything about mod depth accuracy. Siglent looks more right ( image marks A). But also it need note that SA/SVA is near its limits here.
Rigol show, in these images, one non harmonic spur marked B in image.
Of course this kind of extremely tiny test can not at all tell anything like "which one is better". It need thousends times more tests and special tools for do all tests. For normal use for use in this performance class what they are designed for they are like two berries without big differences. Perhaps one have bit better one thing and other bit better other thing. User need select what he like more and what are his need. Only individual user can + and - with his own personal needs and what hive think more important than other thing.
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1039054;image)
Nearly like two strawberries, so different ones and still so same....
-
@rf-loop
Thank you for your observations / comments.
Yes, the simple test plots are not anywhere appropriate to draw conclusions of which device is 'better' - like you said there are numerous tests that need to be performed for a true head-to-head and then under 'controlled' conditions with a high capability analyzer - not one like the SVA used here - which is already operating at its limits.
Nevertheless - these simple tests at least show that BOTH instruments are in the 'same class' - what is / was intended when Siglent Released their SSG to compete with the Rigol DSG800 series.
What probably matters more for the forum audience - is the 'value for money' when comparing Rigol to Siglent offerings.
Here, I feel there is significant opportunity to harness some sales on appropriate price positioning and marketing.
I would certainly buy a 3.6GHz DSG - with its small compact size at a lower cost than the equivalent Siglent 3.2GHz offering.
But there will always be the desire for some to acquire good specimen of an 'oldie' - when size and interface dos not matter - but only its capability, they will be better value always.
Your 'off topic' insights are most welcome - its great to hear such valuable observations - and truly appreciate the 'old school' engineering which prevailed in those days. Products were build to last, commercial aspirations were secondary. Today its the reverse - products are built to a price , performance comes second.
-
@rf-loop
Thank you for your observations / comments.
Yes, the simple test plots are not anywhere appropriate to draw conclusions of which device is 'better' - like you said there are numerous tests that need to be performed for a true head-to-head and then under 'controlled' conditions with a high capability analyzer - not one like the SVA used here - which is already operating at its limits.
Nevertheless - these simple tests at least show that BOTH instruments are in the 'same class' - what is / was intended when Siglent Released their SSG to compete with the Rigol DSG800 series.
What probably matters more for the forum audience - is the 'value for money' when comparing Rigol to Siglent offerings.
Here, I feel there is significant opportunity to harness some sales on appropriate price positioning and marketing.
I would certainly buy a 3.6GHz DSG - with its small compact size at a lower cost than the equivalent Siglent 3.2GHz offering.
But there will always be the desire for some to acquire good specimen of an 'oldie' - when size and interface dos not matter - but only its capability, they will be better value always.
Your 'off topic' insights are most welcome - its great to hear such valuable observations - and truly appreciate the 'old school' engineering which prevailed in those days. Products were build to last, commercial aspirations were secondary. Today its the reverse - products are built to a price , performance comes second.
Btw, have you anyhow looked how much they leak overall or least from front panel and specially TFT. Take small antenna and look with SVA/SSA in place what is free from other sources . Note that many things today also may emit Lot of and every small switch mode circuits what are today in every place polluting all places with enormous RF noise and no one care but all manufacturers claim they test and these meet rules and standards. But when I open The Receiver if it is CEI or Racal or Rockwell Collins all I can hear is enormous noise and need very special things for isolate signals from under this noise.
Now signal generators are also made for test many kind of RF circuits. They must not pollute signals! Who have forget and what. If DUT is near generator and you connect then your probe to DUT do you want see example generator TFT signals or DUT generated things with injected signal from signal generator. Think situation where we are working with quite low level signals. I do not have ANY doupt I can do these with example HP8642B etc... but before I take Sig or Rig to my board I need start deeply inspect these things firast before can continue anything or do I need isolate these inside Faraday box.
But so or so.. it is nice to know how these things are... in these "toys" even when they are quite good overall when look also price if compare other New RF sig gens.
-
That's an area where the "big ones" really excel... My R&S SM300 has a wire mesh sandwiched into the perspex/glass(?) cover of the LCD, connected to the frame ground. In general, shielding and filtering is taken far more seriously in this instrument than what I can tell for the Rigol DSG800. Since the only teardown of the SSG3000X on the web by @Defpom stopped short of opening the "RF Voodoo Box", I cannot compare to Siglent's counterpart. But I guess that it will be on par with Rigol's entry level RF signal generator, give or take a little.
Starting off again somewhat off-topic, I found some interesting details about the reference oscillator of the R&S SM300. Since I scored another specimen of these instruments, for even less money than my first, but with more severe problems, I dived more deeply into the details of this generator. Fortunately, I've been able to sort all the issues and found it to be considerably more accurate than my first unit, showing level errors of about +-0.2dB between -30dBm and +13dBm over its whole frequency range, measured with my HP437B + HP8482A power meter (the first one will have to undergo some surgery again with the lessons learned on the current one... ;)). Below -30dBm, the power sensor won't work accurately anymore, so here I've got to rely on the accuary of the SM330's low-level attenuator. But I haven't got any doubt that it's fairly accurate in this low range as well.
@rf-loop mentioned that the measurements should (at least initially) be taken with the internal reference oscillator of the instruments that are under scrutiny, and I guess that's a reasonable approach to avoid any potentially additionally introduced artefacts. So what I did is to replicate the 100MHz -70dBm carrier, 60Hz 0.1% amplitude modulation test (as suggested by @rf-loop) that I also ran on the first of my SM300s. The analyzer is an SSA3000X. This time, I ran (carrier frequencies superimposed):
both instruments running on their internal references - trace A,
both instruments sync'ed to my Rb reference (based on an Efratom LPRO101) --trace B,
the SM300 on its internal reference and the analyzer on the Rb source -- trace C,
the SM300 sync'ed to the Rb source and the SSA3000X running on its internal reference -- trace D.
(https://www.eevblog.com/forum/testgear/siglent-ssg3021x-testing-feedback-and-comments/?action=dlattach;attach=1047430)
The results are quite peculiar: While the external reference has little effect on the analyzer (except for the increased absolute frequency accuracy), the SM300 shows considerably less phase noise/modulation very close to the carrier while the artefacts @ ~22 Hz off the carrier appear to be a result either of the Rb source or the PLL circuitry inside the SM300. I didn't test higher signal levels which may show more severe effect on the analyzer, may add that later, though.
So the SM300 definitely gains some accuracy phase-noise-wise if a decent external reference clock is used (at the expense of a very small amount of amplitude modulation @ ~22 Hz) while it's not really worth worrying in case of the SSA3000X.
I assume that both Siglent's and Rigol's entry level RF signal generators won't benefit that much from a high quality external reference signal phase-noise-wise.
Recently, I did some interesting experiments with an HP 53320A MDA as an analyzer which resolves the near-carrier range even better, but I'ld like to get the 53305A software package running on an ancient PC setup to provide some information that's better comparable to the readings of an SA.
We may also consider merging the information on different "affordable" RF signal generators into one thread since right now, comparing the instruments is a little difficult, even though there's a lot of cross-information in the corresponding threads already. But I'll leave that as a suggestion to @noreply who put most effort into these threads.
Cheers,
Thomas